The hail comes through Aurora a little after nine on a Tuesday night. Golf-ball size, the kind that dents car hoods and strips shingles in sheets. By morning, driveways across the east metro are littered with asphalt granules, and homeowners are standing in their yards looking up at damage they don't know how to price. A decade ago, every one of them would have opened a browser and started scrolling. Today a large share of them open an app and type a question instead. "My roof got hit by hail in Aurora last night, who's a good roofer to call?" And in the second it takes an AI assistant to answer, the entire local roofing market gets sorted into two groups: the companies it names, and everyone else.
If you run a roofing company, that second group is where years of your work can quietly go to die.
The storm doesn't create demand evenly. It concentrates it.
Here's what makes a storm different from an ordinary Tuesday. On a normal day, roof-related questions trickle in. Someone has a slow leak. Someone's selling a house and wants an inspection. The volume is low, the urgency is low, and the person asking has time to compare several options before they commit. A storm detonates that pattern. In the forty-eight hours after a hail event, thousands of homeowners in the same metro ask nearly identical questions at nearly the same time, and almost all of them are urgent. Tarps are going up. Insurance adjusters are being called. Nobody is leisurely comparing.
That surge doesn't spread itself across every qualified roofer in town. It concentrates onto whoever the assistant already names. This is the part worth sitting with, because it runs counter to how people assume demand works. You might think a flood of new customers would lift all boats, that there'd be so much business after a storm that even the un-named companies catch the overflow. The opposite happens. When demand spikes and urgency is high, the asking behavior gets simpler, not more thorough. A calm homeowner on a normal day might ask follow-up questions, check a couple of names against each other, dig a little. A homeowner with water coming through the ceiling takes the first credible answer and calls it. The spike compresses the decision. And a compressed decision rewards whoever is already in the model's mouth.
So the gap between named and unnamed, which costs you a trickle of business on a quiet week, becomes a chasm in the days when the most business of the year is moving. The storm doesn't level the field. It widens the exact gap you can least afford.
Why the model names one roofer and not the next
An AI assistant isn't ranking a list of ten and showing you the top of it. It's composing an answer, and to put a company's name into that answer it has to be reasonably sure the company is real, local, relevant, and not about to embarrass it. Where does that confidence come from?
These systems don't take your website's word for things. A claim that lives only on your own site — "Aurora's most trusted storm-damage roofer since 2004" — is just an assertion, and the model knows it. What moves the needle is agreement across sources that don't depend on you. When your company name, your service area, and the fact that you handle hail and storm work show up described the same way in several independent places, the model treats that consistency as evidence. It has effectively heard the same thing from multiple witnesses, and multiple agreeing witnesses is how trust gets built, by people and by machines. A model that finds that kind of corroboration grows confident enough to say your name out loud. A model that finds thin, contradictory, or purely self-reported information gets cautious, and a cautious model hands the question to a competitor it can describe without hedging.
That's the mechanism underneath almost every "why didn't it name us" story. Not malice. Not a mysterious algorithm playing favorites. Just a confidence threshold that a well-corroborated company clears and an invisible one doesn't.
Take Yelp as a worked example of how this plays out, because it illustrates the point cleanly. A roofing company's Yelp presence isn't valuable to an AI the way a backlink was valuable to old search engines. Its value is as a corroborating witness. When an assistant is assembling an answer about roofers in Aurora, a Yelp listing is one of the independent places it can check your existence against — does the company described there match the company described on its own site, in the same city, offering the same storm work, consistently named? If the answer is yes, that's a vote toward confidence. If your Yelp listing calls you one thing and lists a service area that doesn't quite line up with your site, or if it barely exists, the model notices the seam and leans away from you. What matters isn't that Yelp is some magic ingredient. It's that Yelp is one of the independent corners of the web where your story can either confirm itself or quietly contradict itself, and the model is reading for exactly that.
Notice what that does and doesn't tell you. It doesn't tell you to go "do Yelp" as step one of a plan. It tells you something about how the whole picture gets read: as a question of whether your story holds together when checked against sources you don't control.
The competitor who's already the answer
Walk the scenario forward. The hail hit Tuesday night. By Wednesday afternoon, a homeowner on South Chambers Road asks Gemini who to call, and Gemini names a roofing company three miles away — one that does roughly the same quality of work you do, charges roughly what you charge, has been in business about as long. The only real difference is that the model is confident about them and uncertain about you. That homeowner calls the named company. They never type your name because they were never given it. You don't lose that job in a bidding war. You lose it before you know it exists.
Now multiply that by every urgent query in the metro over seventy-two hours. The named competitor isn't just getting more leads than you this week. They're getting the surge, the once-or-twice-a-year flood of high-value insurance-backed roof replacements that can define a company's entire year. And here's the compounding part. As they book those jobs, do that work, and get mentioned in more places, the corroboration that made them the named answer gets stronger. The storm that should have been shared across the market instead pours into the companies already named and makes them harder to unseat. Next storm, the gap is wider still.
This is the loss that doesn't announce itself. There's no dashboard that shows you the customer you never heard from. A real homeowner with a real insurance claim and a real roof to replace stood in their yard, asked a machine for help, and was handed someone else's name. That person was yours to earn, and you'll never know they were there.
Looking before the next front moves through
The uncomfortable truth is that you can't fix what you can't see, and almost no roofer has actually watched what the assistants say about their company. You can start by asking yourself, the way a frightened homeowner would — open ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Grok, and Copilot and ask each one who to call for hail damage in your city. Read who gets named. Read whether you do. It's a sobering ten minutes, and it varies by session and platform enough that one good result shouldn't reassure you and one bad one shouldn't be the whole story.
That's the limit of doing it by hand, and it's why RankCommander exists. We check how you're named across all seven assistants, systematically, and keep checking as the models update and the market shifts. Our AI Visibility Index draws on tens of thousands of real AI answers evaluated and still counting, which is how we can tell you not just whether you're named today but how your market behaves, which verticals and metros get surfaced and which get skipped. You can see the methodology and the real, disclosed-sample-size benchmarks at the AI Visibility Index. The scan looks at where your story holds together across independent sources and where it frays, because that's the seam the models read for.
What actually separates the roofers who get named from the ones who don't is rarely the quality of their work. It's whether, on the morning after the hail, a machine can say their name with confidence. You've spent years building a company that deserves to be the answer. The question is whether the assistants know it yet, and whether a competitor three miles away is already standing in the spot that should be yours. The next storm is coming on its own schedule, not yours. Run your scan now and find out who the AI names before the hail does it for you.